Tears of wine

Why does a ring of droplets climb the inside of a wine glass and run back down with nobody disturbing it?

Swirl a glass of wine, set it down, and a clear band climbs the wall above the surface. It gathers into a rim, the rim beads, and the beads run back. Nobody is touching the glass, the room is still. Something is lifting liquid several centimetres against gravity, continuously.

The reflex answer is capillary action — the same thing that pulls water up a narrow tube. But that answer has a cheap test attached, and it fails it. Swirl plain water in the same glass and you get a wet band that drains and stops: no rim, no beads, no recurrence. Whatever is happening needs the wine to be a mixture, and any explanation that would work equally well for water is the wrong explanation.

So what does a mixture have that a pure liquid does not? Two things, and both matter here. Its components evaporate at different rates, and they have different surface tensions.

Take the second first. Surface tension is a force per unit length pulling along a liquid surface, and the values differ sharply: water is about 72 millinewtons per metre at room temperature, ethanol about 22 — both standard recalled figures, but the gap is the point. A wine sits between, and its surface tension depends steeply on how much alcohol is in it.

Now the first. Ethanol is the more volatile component, so it leaves faster than water does. In the bulk that hardly registers — there is a lot of wine under a little surface. But the film smeared up the wall by swirling is perhaps tens of microns thick, so its surface-to-volume ratio is enormous, and preferential evaporation strips alcohol out of it in seconds.

Follow the consequence. The film is now alcohol-poor, so its surface tension is higher than that of the wine below. And here is the step that does all the work: a surface whose tension varies from place to place is not in balance. Tension pulls along the surface in both directions; where it is stronger on one side than the other, the difference is a net tangential force, and it drags the liquid underneath along with it. The film's depleted surface pulls harder than the bulk's, so it hauls wine up the wall. This is Marangoni flow, and it is a gradient effect — a uniformly high surface tension would do nothing at all.

Can we check that the numbers are even plausible? Balance the pull against the weight it lifts. A tension difference Δσ acting along the surface can hold up a film of thickness h raised to a height H when Δσ is about equal to the weight per unit width, ρ*g*·h·H. Rearranged, h ≈ Δσ / (ρ*g H*). Suppose depletion buys a difference of 10 mN/m — a modest fraction of the water-ethanol gap, assumed rather than measured — and the band climbs 20 mm. Then h ≈ 0.01 / (1000 × 9.8 × 0.02) ≈ 5 × 10⁻⁵ m: fifty microns. That is a film thin enough to be nearly invisible and thick enough to bead, which is exactly what you see. The mechanism is not being asked to do anything extravagant.

That also tells you why the tears fall. The climbing flow keeps delivering liquid to the top of the band, where it piles into a ridge heavier per unit width than the film feeding it; once its weight exceeds what the gradient can support, it breaks into drops that run back down. How the ridge picks its droplet spacing is still argued over — descriptions in terms of a ridge instability and of a travelling shock in the film both have support — so treat the beading as the unsettled part of an otherwise settled story.

Which suggests the test the whole account stands on. Everything above depends on evaporation continuing. Cover the glass and let the air above saturate with alcohol vapour, and the gradient has nothing to renew it. The prediction is that the tears slow and stop while the wine is unchanged — and they do. If tears persisted undiminished in a sealed, saturated glass, this explanation would be finished. So would it if plain water in a scrupulously clean glass produced them.

Three refinements connect those steps.

The first is that this is not capillary rise wearing a costume. Capillary rise is an equilibrium: water climbs a tube to a definite height and stops there, and it happens to pure liquids. Tears are a steady flow maintained by continuous evaporation — a system held away from equilibrium, which is why it cycles rather than settling. Remove the driving and it does not stall at some raised height; it drains away.

The second is what is actually in the drops. It is tempting to say the alcohol climbs, or that the tears are concentrated spirit. The opposite is true: the film climbs because it has lost alcohol, so the tears are alcohol-depleted wine returning to the glass. What moves upward is liquid; what moves outward, into the air, is the alcohol.

The third is about gravity's role. Gravity is not merely the obstacle — it is half the loop. Without it nothing would drain back, nothing would remix, and the whole thing would run once and stop.

• Why a drop of soap in a dish of water makes floating pepper flee outward, which is the same gradient acting in a single burst.

Every term the collection defines is gathered in the glossary.